VeriStruct: AI-assisted Automated Verification of Data-Structure Modules in Verus

2025-10-28

AI Summary

Large language models (LLMs) frequently misinterpret Verus’s annotation syntax and verification semantics when verifying Rust data structure modules, hindering automated formal verification. Method: This paper proposes a module-level AI-assisted verification approach. Its core innovations are: (1) a syntax-guided hierarchical prompting mechanism that explicitly encodes Verus type invariants, specifications, and proof structures; and (2) an error-feedback-driven automatic repair phase that iteratively corrects syntactic and semantic errors in LLM-generated annotations. Contribution/Results: The method elevates verification granularity from individual functions to complete data structure modules. Evaluated on 11 representative Rust modules, it achieves full-module verification success on 10 modules and verifies 128 out of 129 functions—yielding an overall success rate of 99.2%. This significantly advances the automation and reliability of formal verification within the Verus framework.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness
• Computer Vision: Large Vision Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

We introduce VeriStruct, a novel framework that extends AI-assisted automated verification from single functions to more complex data structure modules in Verus. VeriStruct employs a planner module to orchestrate the systematic generation of abstractions, type invariants, specifications, and proof code. To address the challenge that LLMs often misunderstand Verus' annotation syntax and verification-specific semantics, VeriStruct embeds syntax guidance within prompts and includes a repair stage to automatically correct annotation errors. In an evaluation on eleven Rust data structure modules, VeriStruct succeeds on ten of the eleven, successfully verifying 128 out of 129 functions (99.2%) in total. These results represent an important step toward the goal of automatic AI-assisted formal verification.

Problem

• Extends AI verification from functions to complex data structures
• Automates generation of abstractions, invariants, and proof code
• Corrects LLM misunderstandings of verification syntax and semantics

Innovation

• Extends AI verification to complex data structure modules
• Uses planner to generate abstractions and proof code
• Embeds syntax guidance and repair for annotation errors